E81 Transcript
A couple weeks ago I wrote a paragraph or two in my weekly newsletter called "Cat Out of Bag." A reveal, more or less, as to why I put typos in my content and the resulting email replies. All of it. I said the typos are staying because it is the only sure bet that I wrote it not some form of AI that has access to my platforms.
Now that is a loaded topic, I know. And by the end of that piece I told you flat out that this topic was too big for a brief newsletter mention. That it deserved a real conversation - twenty-five to thirty minutes, the whole thing, not a paragraph hiding inside a newsletter. I said a podcast episode was coming.
Well, here we are.
This is that episode.
So, we’re going to talk about it. Where I use AI. Where I refuse to use it. And exactly why I draw the line where I draw it. No hedging today. No polite public version of this. The real one.
If you've been waiting to hear where I land on this topic – for real, not the PC version – this is it. Stay tuned.
Let me be up front, plain and simple, so there's no confusion about where this episode is going.
I use AI. I use it consistently in the business side of what I do. Research. Organizing scattered thoughts. Brainstorming. Rough outlines. A sounding board when I need to think through something. It is woven into how I run the business every single day.
And I never use AI. I never use it for the actual creative work. Not the artwork. Not the pattern design. Not the craft itself. None of it. Every piece of that comes from my own mind and my own hands, the same way it always has. No AI creative plans, No AI drawings, No AI compositions. No AI photographs. I know of no AI programs that can sew a straight seam on silk or glue down a piece of fiber.
And here’s where I stand on it:
I think of it as a virtual assistant. Not a clone of my ability.
A tool, not a replacement for talent.
And I want to say something else before I go any further, because I think this really matters. There is real disdain out there right now for people using AI to look more talented than they actually are. I understand that disdain. I share a good amount of it.
This episode is not a defense of that. Far from it! It's the opposite, actually. By the end of this I think you'll see exactly why that kind of shortcut doesn't hold up - and why it was never really about the tool at all.
I've lived through a version of this conversation once already. So let me tell you about that, because I think it tells you almost everything you need to know about how this one is going to go too.
I went through design and art school before the advent of computers and computer aided design. I learned the drawing-by-hand method. Pencils, pens, erasers. I learned to make patterns by hand. Heavy paper. Ruler. A big table. Enormous paper shears. That was my entire toolkit for years. Not a mouse or a stylus or a tablet.
And then computer-aided design systems started showing up in the industry. CAD systems they called them. And I want to be honest about what that moment actually felt like, because it wasn't calm and it wasn't unanimously good.
There was real fear. Extreme resistance. Some of the old school patternmakers and designers I worked alongside saw it as a cheat. As if pulling a line on a screen instead of a table somehow didn't count as real skill anymore.
And on the other side, there were new students coming up who thought the opposite thing and were just as wrong. They thought CAD was going to be their shortcut. That they could ride the software straight to the top without putting in the years of sustained practice the rest of us had put in.
Neither one of those things turned out to be true.
The designers who were genuinely good - and I mean the ones who could already draw, who already understood proportion and movement and how fabric behaves on a body - they learned the new tool and got faster. That's it. That's the whole effect. Faster. Not different. Not better at the actual skill. Just faster at expressing a skill they already had.
And the students who thought the computer would carry them? It didn't. I heard it constantly back then - I'm going to make it because I can use a computer to do the work. And then reality caught up with them, because using the software still required knowing which line was wrong. Which shape needed to change to make the thing actually sew together. They didn't have that knowledge yet and it became obvious very quickly. The tool had nothing to hide behind for very long.
Here's the part that matters most from where I'm sitting today. When CAD entered my world, I was already excellent at design and pattern making. I already knew exactly what a good armhole looked like and precisely what needed to shift to make the fabric work. All I had to learn was a new kind of pencil. The new graduates coming out of school had to learn the drawing skills, the pattern skills and the computer skills at the same time - and that gap showed up fast and hard. You see, at first it was the fear of something new and not yet understood. It was the thought that all you had to do was type into the computer that you wanted to change the straight skirt to a circle skirt, and poof it would appear. Not so. You had to actually work the shapes and know what was right and what was wrong. A lot of very bad patterns were released until finally the users realized, okay-this isn’t as easy as I thought. It’s not a substitute for knowing what I doing, or being good at something.
I am watching the same exact pattern play out right now with AI. Different tool. Same underlying truth. I’ll explain why in a minute. The difference is that it’s just moving faster and louder this time.
So let me give you the piece of this that I think explains everything else, if you only remember one thing from this whole episode.
It’s not about the tool. It never was about the tool itself.
Think about money for a second. A lot of money does not make a person bad. The money isn't the problem. What money actually does is magnify the type of person someone already was before they had it. Give a lot of money to someone who was already unscrupulous, and they get more unscrupulous - they have more resources to be unscrupulous with now. Give a lot of money to someone who was already kind and generous, and they get to be more kind and more generous, at a bigger scale than they could manage before.
Any new tool does exactly that. CAD did it. Good designers became faster at being good designers. Bad designers got to expose their bad things more easily. AI is doing the same thing right now. It doesn't instill character. It doesn't remove it either. It just shines a magnifying glass on whatever was already there and makes it bigger and more visible to everyone watching.
So, let's talk about the disdain for such things, because I promised you I would.
There will always be people trying to short-circuit the system. Someone generating "their" (in air quotes) artwork with AI and presenting it as original, handmade work. Someone writing content with a tool and presenting it as a deeply personal experience that never actually happened to them. Someone building a portfolio or a skill level on screen that doesn't exist anywhere else in their life.
Some of those people will get exactly what they wanted. For a while.
And then it falls apart. Every time. Dishonesty has a shelf life and it's shorter than people think when they're standing inside the early success of it. The consequences level out eventually. They always do. This isn't a hope I'm holding onto - it's the same pattern I already watched happen with CAD, just running hotter and faster this time around because the tool is louder and the audience is bigger.
Think Milli Vanilli. The climb can be real. People will cheer. And then the fall, when it comes, is hard, and uncomfortable, and more than a little bit funny to watch from the outside once everyone realizes what they were actually looking at.
Now I want to get specific, because a vague claim about "using AI responsibly" doesn't actually mean anything to anyone. Let me tell you exactly what this looks like in my business, day to day.
I use it for research - finding information that would otherwise take me weeks to dig up on my own, the way a research assistant would do it for me.
I use it to organize scattered thinking. I will dump a huge pile of unstructured thoughts into it (I’m talking two-three and even four single spaced pages worth) and ask it to sort them into categories, find the throughline I couldn't see myself, group things by type.
I use it as a sounding board. Really - like a friend at a coffee shop, except without burdening an actual friend with the details of running a business they don't run themselves. My real friends are wonderful and most of them have absolutely no context for what a launch week feels like or why a pricing decision is keeping me up at night. This gives me somewhere to think out loud and vent about that without making it someone else's burden.
I use it for task and time organization - loading in everything I need to get done and asking it to batch things by type or by the kind of energy each task requires, so I can use my actual time better.
I use it for brainstorming and for finding a word that's sitting just out of reach. I'll describe a feeling or a half-formed thought I can't quite articulate and ask for three or four ways to say it, and usually one of them is the one I was actually reaching for.
I use it to build SOPs (that’s standard operating procedures) so that when something comes up that I need to tackle, I have a written account of how I do things here at VLS. That can be everything from the standard replies to speaking inquiries to the various show set up and break down routines.
I use it for a rough first outline when I'm building a workshop or a class - just to get a skeleton on the page before I start doing the actual work of building real content into it.
There are specific things I have to teach it so that it knows how to organize my time and my space and my thoughts so everything remains congruent with my meaning and porposes.
You can see that it truly can be a life saver in the business sense.
BUT - here's what never happens, not once, not ever. The actual creative work. No artwork. No pattern design. No finished craft of any kind. Every single piece of business writing that comes out of this process gets rewritten and reworked by hand - often several times - until it’s just like me and not like an average of every business on the internet. Nothing gets posted, published, or sent if I didn’t get deep into it myself.
Now I want to spend some time on something I think is actually the most important part of this whole conversation. Because I think most people arguing about AI and creativity right now are arguing the wrong thing.
The question that actually matters isn't whether AI can produce something that looks creative. It clearly can, in a narrow sense. The real question is what specifically guarantees that human creativity and lived experience hold ground that nothing generated can ever actually occupy. And I think there are some very specific, very concrete answers to that.
Start with this. A generated output has no stake in the outcome. When I make something, there is a version of me inside it that genuinely wants it to be right - wants it to fit, wants it to hold up, wants the person wearing it or viewing it or living with it to feel something good. That caring shapes a hundred small decisions along the way in ways I'm not even fully conscious of while I'm making them. An AI generator has no investment in whether what it produces actually works for anyone. It will generate the next thing exactly as swiftly whether the last one was a triumph or a disaster. Caring isn't a feature you can bolt onto intelligence. It's a completely separate thing, and it's threaded through every decision a real maker makes.
Next - no generated output has a body that has made mistakes and learned from them. My hands know things that my conscious mind forgot it ever taught them. Yours do too! The exact pressure for a particular seam. What a fabric feels like in the seconds right before it's about to pucker under the needle. That knowledge isn't stored anywhere as information you could write down and hand to a machine. It's stored as experience, in an actual body, that got it wrong hundreds of times on the way to getting it right. Nothing generated has ever threaded a needle badly and felt the specific small frustration that taught it something permanent.
Nothing generated wants something for reasons that came out of an actual specific life. My work carries the real weight of a real December finalizing my mother's estate. A real fear during COVID, starting over with no map. A real decision about which kind of work to keep saying yes to and which kind to walk away from. None of that is decoration sitting on top of the work afterward. It's structurally why specific choices got made in the first place. Lived experience is causally connected to the actual decisions inside the work. AI can describe grief, or risk, or starting over - because it's read millions of words other people wrote about those exact things. But it's only ever working from what's already been said about them. It cannot make a single decision that came from an actual Tuesday that actually happened to it.
Next - and this is the one I think people miss the most. SO listen carefully here - These systems, these AI systems – all of them - are built to regress toward the average. Let me say that again - These systems are built to revert toward the average. That’s NOT a limitation that a better version will eventually fix. It’s the actual mechanism underneath how they work. A system trained to predict the most statistically probable next output is, by definition, suppressing the unlikely choice. The idiosyncratic one. The slightly weird, personal, nobody-else-would-have-done-it-that-way choice. (BIG SMILE HERE) And that exact kind of choice is usually the thing that makes a piece of work distinctly someone's and not generic. And you can tell immediately when you see it or hear it or read it.
Excellence, if you want a real definition of it, is the deviation from average made by someone who cared enough and lived enough to make a non-obvious choice on purpose. That is the literal opposite of what these systems are built to produce. Okay? They are not built to produce excellence. They are built to find the average between the worst and the best- and that’s all.
And last - nothing generated can feel the gap. We spent an entire episode a few weeks ago on exactly this. The internal signal. The self-reckoning. The one bad dress early in my career that turned into a permanent lesson because I genuinely felt the distance between what I'd committed to and what I'd actually delivered. That signal only exists because something is at stake personally. A generated output can be told a result failed. It cannot feel that it failed. And that feeling - that specific, uncomfortable, motivating feeling - is the entire engine behind everything that drives a real maker toward a higher standard over time. Ai can’t and won’t do that for itself. Take that away and there's no engine left. Just average output.
Put all of that together and here's where I land. There is no continuity of self across generated work, because there's no self there in the first place for a throughline to belong. My body of work has an actual story running through it because I am one continuous person who has been making increasingly informed decisions for many decades. That throughline cannot exist without someone actually there to carry it.
So here’s where I want to address the “What about if…” scenarios, because I know we’ve all seen some “successes” out there completely dependent on AI.
Right now there are people building entire libraries of AI generated songs, AI generated books, AI generated everything, and some of them are seeing real success with it. They’re getting downloads, sales and attention. And I understand why that looks like proof the tool works just fine without a real person behind it.
But think about what's actually happening underneath that success. Every one of those systems is producing the average of everything it's already seen. And the more people use it the same way, the more that average gets fed back into itself. The next version learns from the last version's output. Which was already the average of the version before that.
That doesn't make things better – that doesn’t make things more original over time. It makes them sound and look more like each other. The songs start to share the same shape. The books start to share the same rhythm. The thing that felt fresh the first fifty times you encountered it starts to feel exactly like the last thing you heard, because in a real sense, it is. It IS the average of the last fifty things you heard.
And here's the part that should worry anyone leaning entirely on the tool for creative reasons. If a system can produce that average output for one person, it can produce the exact same kind of output for a million other people doing the same thing. There's nothing in that pile of generated work that's actually theirs. It's just one version of the same average, with their name attached to it instead of someone else's.
The sameness doesn't shrink over time. It grows. That's the trajectory these systems are actually on, by design, not by accident.
So a reckoning is coming for the people who built their whole approach on the tool doing the work for them. Not because anyone's going to call them out publicly, necessarily - though sometimes that happens too. The reckoning is quieter and more thorough than that. It's the moment the output stops standing out at all, because everyone's output started looking the same the moment everyone started asking the tool the same kinds of questions.
And that's really the whole thing in one sentence, if you want it. The answer is only ever as good as the question behind it. A generator with no lived experience, no personal stake, and no actual taste of its own can only ever hand you back a version of what you asked for. If the question came from somewhere shallow, the answer will be shallow too, no matter how polished it sounds on the surface. The people who are going to be fine on the other side of this are the ones who were already asking better questions to begin with - because they had something real underneath the question.
So let me bring this all the way back around to say-
It was never about the tool. CAD didn't create talent and it didn't destroy it either. It revealed who already had it, fairly quickly I might add. AI is doing the exact same thing right now, just faster, and considerably more ‘in public.’
The people trying to shortcut their way into looking talented are going to get some of what they wanted. For a while. And it is not going to last, because it never has, not once, in any version of this story that's come before.
My own practice, one more time so there's no ambiguity about it: I like it and use it fully as a business tool. Zero use as a replacement for the creative work. Even the organizational directives and words it hands back to me, are the product of a brain dump on my end. And it all gets sifted and sorted, picked apart and dissected, rearranged and reorganized. Read, reworked, and scratched up by me before it goes anywhere near anyone else's eyes.
And that typo I mentioned at the very beginning of this episode - the one that started this whole thing as a throwaway joke – I decided I'm keeping it. On purpose, from here on out. Because my fingers are going a mile a minute at the keyboard and I miss a thing every now and then. I run it thru spell check and grammar check, but even that is fallible. A small, slightly imperfect, completely recognizable signature. Proof, in miniature, that there's an actual human hand still on every piece of this.
It connects to everything we've talked about the last few weeks, even though it doesn't look like it on the surface. The same commitment to an honest standard and an honest signal that we covered with the mediocrity flood, with the pressure to lower the bar, with what depletion does when nobody's looking - all of it shows up here too. Just applied to a tool instead of a technique.
Remember, it’s not about the tool. It has never once been about the tool.
Every tool that's ever come along - CAD, social media, AI, whatever comes after AI - does the same single thing. It hands you a magnifying glass and waits. And eventually it shows you exactly who someone already was before they ever picked it up. The patient ones get more patient. The honest ones get more honest. The ones cutting corners get caught cutting bigger onesllll.
You can tell almost everything you need to know about a person by watching what they do with a powerful tool when nobody's checking. Whether they still do the work. Whether they still care if it's actually good, or just whether it looks good long enough to get past someone or get paid. That choice was never going to come from the software. It is always going to come from the person.
So, use the tool with intention. Keep the line exactly where it belongs. And let the work that actually matters keep coming from your own mind and your own hands - because nothing else can make it the way you can, nothing else can be held accountable for it the way you can, and nothing else can reap the long term benefits the way you can either.
I'll see you next week.
